DS2 - ANM 694: WEEK 9

Assignment 9.1: Create Facial Rig and UI

Created the facial rig based on the video demos. Rigged and skinned the head as well as rebuilt and set up the eyes.

Planning to circle back and push the blendshapes a bit this week to increase potential range of some expressions per feedback.

I have run into an issue with the setup as laid out in the videos. The instructions state to parent all the face joints to the head joint, but if I do that, rotating the jaw doesn't take them along and the associated skin weights pull away. Any ideas how to fix this?

Temporarily I have parented the lower lip joints to the jaw joint, but the corners should really split the difference. Did I miss something in the videos to address this?

That 'explosion' issue might be double transforms -- if you parent the controller circles AND any of the joints (rig copy or ones on face) you get twice the rotation of whatever you've parent to (ostensibly the head joint or control.)

Rigging is definitely a challenging business!

INSTRUCTOR FEEDBACK

Very good job with this. Appreciate the progress made for this facial rig. Eye controls work, head rig controls are working but there's a noticeable lack of Facial UI for blendshapes. Also good job with naming each of the animation controls. For any issues with troubleshooting the rigs, joints and parenting can also be directed to online Maya workshops where rigging concerns can also be brought up and worked on.

Assignment 9.2: Chapter 12 of Stop Staring.

Discuss what you think about blendshapes and facial rigs. Do you find that you prefer one over the other? What would you like to see in the future development of facial manipulation technology?

In looking at blendshapes versus facial rigs, there are definitely pros and cons to both systems. Blendshapes offer much more specific control in terms of being able to sculpt very detailed and accurate expressions involving wrinkles and dramatic shape changes. Facial rigs offer much better in-between and flexible control of facial structures. One issue with blendshapes can be getting good transitional arcs between set shapes; when moving from one head shape to the next, the vertices move in a straight path from the start position to the end position. This isn't very appealing. If one were animating from "smile" to "frown," there isn't a good way to delay or offset the movement of one part of the face during the animation. Facial rigs offer very granular animation control of features, but skinned meshes will not have the detail or accuracy of sculpted ones.

A combination of both techniques would be optimal if possible. This way, the facial rig could modulate the skin during in-betweens, allow for better motion arcs and help to mitigate the very linear quality of transition inherent in blendshapes. A facial rig would also add variation to the motions of the face, allowing for more and different facial poses than the shape library provides.

For future technologies, I think the involvement of muscular simulation would greatly benefit facial animation. Instead of "sculpting" blendshapes, one could rig the face in a way that realistically moved the skin and flesh as the muscles moved for their purposes.

INSTRUCTOR FEEDBACK

Great points,

I'll also mention from a historical perspective the different approaches of facial expressions with rigs vs shapes vs simulation has been explored for many years and decades in CG production.

Decades back in the 1990's the software R&D team at PDI before they were acquired by DreamWorks Animation looked at shapes vs muscle sim. One software engineer was advocating shapes and his idea was shot down for a muscle simulation solution. That engineer would leave PDI and take his approach to ILM which was about to start production on Dragonheart. This was pre-Maya and the ILM's CG department would develop a proprietary facial animation system based on shapes somewhat akin to blendshapes in Maya.

I mention this to illustrate two different facilities approach to facial animation one being a artist (modeling/sculpting) driven process with shapes at ILM doing photorealistic CG characters and another studio that would be known more for cartoony CG characters that approached it as a technical muscle simulation process for animated features. There is also other ideas like clusters organizing a set of points to manipulate instead of shapes as well.